On December 8, 2025, Integral AI held a focused in-person briefing at Google for Startups Campus Tokyo to provide technical context on its newly announced AGI‑capable model. The session offered a first look at the architecture and safety principles behind the system and illustrated how the model autonomously learns new skills safely and efficiently. Attendees heard from CEO and co‑founder Dr. Jad Tarifi, who explained the model’s ability to grow, plan and act as a unified system. Professor Tetsuya Ogata of Waseda University joined Jad for a fireside discussion on the broader implications for robotics and embodied AGI.

The event coincided with Integral AI’s public announcement that it has successfully tested the world’s first AGI‑capable model. The company defined three qualifiers for AGI readiness—autonomous skill learning, safe and reliable mastery and energy efficiency—and reported that its model meets these benchmarks, enabling robots to learn new tasks in real‑world settings without human supervision. The full press release is available in English and Japanese.

A short AGI sandbox demonstration video is also available to showcase these capabilities: